Abstract
INTRODUCTION: Bone lesions are sometimes detected on computed tomography studies, and biopsies are performed to evaluate whether these are malignant. The aim of the study is to evaluate whether chemical-shift imaging (CSI) and diffusion-weighted imaging (DWI) magnetic resonance imaging (MRI) are more informative than the CT attenuation for predicting malignancy. METHODS: Retrospective analysis of 86 patients who underwent both diagnostic CT, CSI MRI and DWI MRI within 6 weeks prior to bone biopsy at a tertiary care academic institution between 01/01/2010 and 03/01/2020. The CT attenuation, signal intensity on in-phase sequences (SIIP), signal intensity on out-of-phase sequences (SIOP), signal intensity ratio (SIR = SIOP/SIIP) and the apparent diffusion coefficient (ADC) of the lesions over the region of the biopsy tract were measured. RESULTS: A threshold CT attenuation of 157 Hounsfield Units (HU) had a sensitivity of 47.7%, specificity of 83.3% and area under the curve (AUC) of 0.59. A threshold ADC of 793 × 10(-6) mm(2) /s had a sensitivity of 75.8%, specificity of 85.7% and AUC of 0.83 to predict whether a bone biopsy would detect malignancy. A threshold SIR of 0.949 had a sensitivity of 77.8%, specificity of 77.8% and AUC of 0.81 to predict whether a bone biopsy would detect malignancy. ADC (P = 0.029) and SIR (P = 0.009) were significantly better than CT attenuation. There was no predictive difference between SIR and ADC (P = 0.742). CONCLUSIONS: The CT attenuation of a lesion is a poor predictor of malignancy in bone lesions. CSI and DWI are significantly better for predicting malignancy.